"Never believe an alcoholic"? On the validity of self-report measures of alcohol dependence and related constructs

A study of 114 male and female alcoholics was conducted to evaluate the validity of information obtained by means of self-report questionnaires. Factor analysis was used to derive composite measures of alcohol dependence, withdrawal symptomatology, pathological intoxication, and alcoholic psychosis. Validation was conducted by comparing these measures with independent and external criteria; that is drinking estimates made by collateral informants, measures of general alcohol involvement, and drinking behavior 6 months after treatment. The results support each type of validity, and thereby contradict the assumption that the alcoholics' self-reports were not accurate. The question of unreliable self-report data due to the demand characteristics of the situation under which the information is obtained is also addressed.